Project Description

advertisement

Project Description

Axial fans have high performance airflows, relatively good efficiency, and have an axial airflow pattern. Their major advantage is their capacity to produce high axial airflow rates, which make them a perfect choice in cooling applications e.g. cooling of a CPU

Microprocessor.

Selection of an airfoil blade profile which best suits the application is still a challenge, and a lot of research is going on in this field for proper blade profile selection. For this project, we have used NACA-65 blade profile, which is most commonly used in axial fans by Manufacturers. Since these fans operate at high rotational speeds (3450 rpm in our case), high stresses are developed in the fan body due to centrifugal force. The stresses are especially high at the contact area between the fan blade and the hub.

Moreover, the blade section size varies over its length with minimum at the hub and maximum at the blade tip. This makes the joint most probable region for failure. Hence we have concentrated at the hub and blade junction area for stress analysis.

The stresses at the junction vary with the angular speed of the shaft. But as we have concentrated on the static analysis rather than dynamic, we have kept the angular speed of the shaft as constant. Therefore static analysis is carried out by freezing the fan rotation at a specific instant.

1

The cooling fan used for stress analysis has following geometrical specifications

[5]

:

No Parameter

1) Hub Radius

Value

225 mm

2)

3)

4)

6)

Hub Thickness

Blade Length

Airfoil Shape

No of Blades

80 mm

276 mm

NACA- 65 series

7

7)

8)

9)

10)

Rotational Speed 3450 rpm

Blade Angle at the Blade Tip

19 º

Material

Yield Strength (Sy)

Al 6061 T-6 (Heat Treated)

241 N / mm

2

Another important factor for choosing “Axial Fan”, as a model for stress analysis was the effective use of Pro/Mechanica feature called, “Cyclic Symmetry”. For rotating objects having cyclic symmetry (i.e. a certain pattern repeats in the body identically), we can model and analyze only part of the structure with cyclic loads and constraints and still the results obtained are true for the entire structure. This causes considerable saving in CPU time. Thus we have modeled and analyzed a single blade with 1/7 th

portion of the hub and have achieved results valid for the entire fan structure.

2

The Model

As mentioned earlier, we have used the Pro/Mechanica feature “Cyclic Symmetry” for analysis. Since all the blades are identical and also the hub is symmetric, instead of drawing the whole fan structure, we have drawn only a single blade and part of hub over which it fits. Thus the model drawn was actually one seventh of the total fan structure.

The modeling steps have been shown in detail as below.

1) We used “Newton – Millimeter- Second” as our system of units.

2) We used 3D solid as our model type. (We first tried to use shell elements for modeling. But it didn’t work since the geometrical cross section of the blade was nonsymmetric. Pro /E could not locate mid planes for the blade cross-section)

2) For creating the hub, revolve tool was used. We used “front plane” as our sketching plane and “right & top” as reference planes. A rectangle with dimensions (225mm X

80mm) was drawn. Vertical axis was selected as the axis of revolution. The angle of rotation was kept as 51.43º (360 / 7) as mentioned above. The sequence of operations is as below:

Revolve >> Sketch >> Select planes >> Draw vertical axis >> Draw rectangle >>

Dimension it >> Accept >> Angle of rotation = 51.43º >> Okay.

3

3) The next task was to model the fan blade cross-section. We used NACA-65 series airfoil cross-section. We struggled and realized that it was not possible to draw the exact airfoil cross-section manually. After intense search on the internet for the specific airfoil cross section shape, we could access the NACA website to get a file containing the (X, Y, Z) co-ordinates of the points which formed the NACA-65 series blade cross section. We loaded those co-ordinates into an MS excel file.

We created an arbitrary sketch using the “splines” function. We provided coordinate system to the sketch. It produced a *.PTS file. This file stored the (X, Y, Z) co-ordinates of different points of the sketch. We modified this *.PTS file by replacing the current co-ordinates by the co-ordinates of NACA-65 series excel file. Thus a new *.PTS file was created with the required airfoil cross section.

4) To model the fan blade, we used “Sweep Blend” tool. We created datum planes for the blade trajectory and blade sections. The blade sections were created using the coordinates from the excel file. We used the “Data from file” function in Pro/Mechanica for this step. The two blade sections (One at the hub and other at the blade tip) were kept at an angle of 19 º as mentioned in the fan geometry specifications. The steps involved are mentioned as below:

Insert >> Sweep blend >> Protrusion >> Menu manager >> Done

Original trajectory >> Define >> Sketch trajectory >> Font plane >> Okay >> Default

4

References >> Select appropriate references >> Close >> Define trajectory >> Okay >>

Z axis rotation zero

Sketch >> Data from file >> Select the required *.PTS file >> Place the section at the center of the hub >> Keep scale of section as 100 >> Okay >> Z-axis rotation 19º.

Sketch >> Data from file >> Select the required *.PTS file >> Place the section at the center of the hub >> Keep section scale as 140 >> Okay

The model was ready

5) System of units for the model was fixed to “N-mm- S” as below:

Edit >> Setup >> Menu Manager >> Units >> Select “mm-N-s” >> Set >> Done.

5

Analysis

For using cyclic symmetry, we had to change the co-ordinate system from WCS to CSO as follows:

Current system >> Create >> Type-Cylindrical >>

References >> Front-Right-Top planes >> Ok

Application of loads

: We considered the “Centrifugal force ( mrω 2 )

” due to high rpm (3450) as the only major force acting on the structure, in radial direction. For assigning this load we calculated the angular velocity

ω

to be 361.28 (rad/s). The velocity vector was pointing along the fan hub axis (in y direction) because fan is rotating in anticlockwise direction when seen from top. The steps involved are shown as below:

Applications >> Pro/Mechanica >> Units >> Continue >>

Menu Manager>>Structure>>Loads>>New >> Centrifugal.

6

Material Selection :-

We selected wrought Aluminum (Grade T-6) as the material for the entire fan structure.

The steps involved in assigning the material are as follows:

Structure >> Materials >>AL6061 >>Assign part >> Ok

Applying constraints :-

We used two types of constraints, one for cyclic symmetry and the other for hub surfaces.

We applied the cyclic symmetry constraints on the two side faces of the hub which are at an angle of 51.43º. For deflection constraints, we changed the co-ordinate system from

WCS to CSO. Hub surface constraints are applied on the top, bottom as well as on the curved surface of the hub. For constraints in translation R and Z were kept free and Theta was fixed while in rotational constraints all R, Z and Theta were kept fixed. We also fixed the Y-axis of the hub in all translations and rotations. The steps involved are shown below:

Structure >> Constraints >> New >> Cyclic symmetry >> Select first & second surface

>> Ok

7

Constraints >> New >> Surface >> Select top, bottom & curved surfaces >> Select coordinate system CSO >> Translation R&Z Free, Theta fixed >> Rotations R,Z &Theta fixed.

Constraints >> New >> Edge/Curve >> Select Y axis >>All translations & rotations fixed

>> Ok >> Done/Return

8

After applying all the constraints and load the model looked like shown below:

We used rigid connection between the fan blade and hub as demonstrated below:

Structure >> Connections >> Rigid connections >> Select hub & blade junction Edge >>

Ok.

9

Analysis & Studies

For analysis of the model, a new static file was created. The model was run for a quick check. Once this run was complete, then the model was run for multipass adaptive check.

We checked the status of the run in status window for both quick and multipass adaptive check. The multipass adaptive check was also completed and results for the analysis were obtained. It took four hours and ten minutes to complete this run. The steps involved in setting the analysis are listed below:

File >> New static >> Provide name & descriptions >> Select quick check >> Ok >> Run >>

Check the status >> Run completed.

File >> New static >>

Provide name & descriptions >> Select multi-pass adaptive check >> Percentage

Convergence 10 >>Ok

>> Run >> Check the status completed.

>> Run

10

Results

Results were obtained for Von Mises stress and Displacement for the model. We also could observe the animated deflection of the model. Maximum Von mises stress was witnessed at the junction of the fan blade and the hub, as expected. Since the blade thickness was minimum at the ends and maximum in the mid-section, the blade was most likely to fail at the ends of the cross-section at the hub-fan blade junction.

------------------------------------------------------------------------------------------------------------

Pro/MECHANICA STRUCTURE Version 24.8(819)

Summary for Design Study "Finalproject1multi"

Tue Dec 02, 2003 13:10:51

------------------------------------------------------------

Run Settings

Generate elements automatically.

Checking the model after creating elements...

No errors were found in the model.

Pro/MECHANICA STRUCTURE Model Summary

Principal System of Units: millimeter Newton Second (mmNs)

Length: mm

Force: N

Time: sec

Temperature: C

Model Type: Three Dimensional

Points: 146

Edges: 647

Faces: 879

Springs: 0

Masses: 0

Beams: 0

Shells: 0

Solids: 377

Elements: 377

------------------------------------------------------------

Standard Design Study

Static Analysis "Finalproject1multi":

Convergence Method: Multiple-Pass Adaptive

Plotting Grid: 4

Convergence Loop Log: (13:12:18)

>> Pass 1 <<

Calculating Element Equations (13:12:19)

Total Number of Equations: 384

Maximum Edge Order: 1

Solving Equations (13:12:21)

Post-Processing Solution (13:12:22)

Calculating Disp and Stress Results (13:12:22)

Checking Convergence (13:12:25)

Elements Not Converged: 377

Edges Not Converged: 647

Local Disp/Energy Index: 100.0%

Global RMS Stress Index: 100.0%

Resource Check (13:12:26)

Elapsed Time (sec): 96.92

CPU Time (sec): 87.05

Memory Usage (kb): 185060

Wrk Dir Dsk Usage (kb): 2049

11

>> Pass 2 <<

Calculating Element Equations (13:12:28)

Total Number of Equations: 2214

Maximum Edge Order: 2

Solving Equations (13:12:29)

Post-Processing Solution (13:12:30)

Calculating Disp and Stress Results (13:12:31)

Checking Convergence (13:12:34)

Elements Not Converged: 128

Edges Not Converged: 496

Local Disp/Energy Index: 100.0%

Global RMS Stress Index: 34.0%

Resource Check (13:12:35)

Elapsed Time (sec): 106.22

CPU Time (sec): 91.29

Memory Usage (kb): 186148

Wrk Dir Dsk Usage (kb): 2071

|

|

|

|

>> Pass 8 <<

Calculating Element Equations (14:03:47)

Total Number of Equations: 43517

Maximum Edge Order: 8

Solving Equations (14:06:48)

Post-Processing Solution (14:59:13)

Calculating Disp and Stress Results (15:00:39)

Checking Convergence (15:01:36)

Elements Not Converged: 3

Edges Not Converged: 0

Local Disp/Energy Index: 22.4%

Global RMS Stress Index: 14.2%

Resource Check (15:04:28)

Elapsed Time (sec): 6819.17

CPU Time (sec): 6193.79

Memory Usage (kb): 439039

Wrk Dir Dsk Usage (kb): 501225

>> Pass 9 <<

Calculating Element Equations (15:23:07)

Total Number of Equations: 54239

Maximum Edge Order: 9

Solving Equations (15:29:30)

Post-Processing Solution (17:05:55)

Calculating Disp and Stress Results (17:09:05)

Checking Convergence (17:10:37)

Elements Not Converged: 3

Edges Not Converged: 0

Local Disp/Energy Index: 22.2%

Global RMS Stress Index: 13.1%

RMS Stress Error Estimates:

Load Set Stress Error % of Max Prin Str

---------------- ------------ -----------------

LoadSet1 2.71e+00 0.3% of 7.86e+02

** Warning: Convergence was not obtained because the maximum

polynomial order of 9 was reached.

12

Resource Check (17:17:39)

Elapsed Time (sec): 14809.08

CPU Time (sec): 13325.08

Memory Usage (kb): 569259

Wrk Dir Dsk Usage (kb): 801019

Total Mass of Model: 5.743880e-03

Total Cost of Model: 0.000000e+00

Measures:

Name Value Convergence

-------------- ------------- -----------

max_beam_bending: 0.000000e+00 0.0%

max_beam_tensile: 0.000000e+00 0.0%

max_beam_torsion: 0.000000e+00 0.0%

max_beam_total: 0.000000e+00 0.0%

max_disp_mag: 1.863701e+00 0.2%

max_disp_x: 3.208244e-01 0.2%

max_disp_y: -1.833184e+00 0.2%

max_disp_z: -1.393368e-01 0.3%

max_prin_mag: -7.862955e+02 13.7%

max_rot_mag: 0.000000e+00 0.0%

max_rot_x: 0.000000e+00 0.0%

max_rot_y: 0.000000e+00 0.0%

max_rot_z: 0.000000e+00 0.0%

max_stress_prin: 3.768182e+02 25.0%

max_stress_vm: 5.632143e+02 10.7%

max_stress_xx: -5.907527e+02 8.2%

max_stress_xy: -2.308429e+02 23.7%

max_stress_xz: -1.566107e+02 12.8%

max_stress_yy: -3.936288e+02 18.3%

max_stress_yz: -1.753765e+02 24.7%

max_stress_zz: -3.576849e+02 17.9%

min_stress_prin: -7.862955e+02 13.7%

strain_energy: 2.066531e+03 0.8%

Analysis "Finalproject1multi" Completed (17:17:52)

------------------------------------------------------------

Memory and Disk Usage:

Machine Type: Windows NT/x86

RAM Allocation for Solver (megabytes): 128.0

Total Elapsed Time (seconds): 14831.94

Total CPU Time (seconds): 13325.76

Maximum Memory Usage (kilobytes): 569259

Working Directory Disk Usage (kilobytes): 801019

Results Directory Size (kilobytes):77130 .\Finalproject1multi

Maximum Data Base Working File Sizes (kilobytes):

540672 .\Finalproject1multi.tmp\kblk1.bas

185344 .\Finalproject1multi.tmp\kel1.bas

21504 .\Finalproject1multi.tmp\oel1.bas

------------------------------------------------------------

Run Completed

Tue Dec 02, 2003 17:18:02

------------------------------------------------------------

13

Maximum Von-Mises stress was found to be = 563.21 N / mm

2

Hence the Factor of safety (FS) was calculated as,

FS = (Yield strength of the material) / (Maximum Von-Mises stress in the material)

FS = (241 / 563.21) = 0.43.

Comment on results : - As we obtained our Factor of safety below 1, it was clear that the part was going to fail. So it was necessary to change our design parameters to keep our factor of safety above 1.5. So rather than playing with model with trial and error method for changing design parameters, we did optimization for minimum mass but kept Von

Mises stress limited to 241 N/mm 2 .

Structure >> Mecstruct >> Results >> Result window definition >> Select the proper

Design Study File.

To display Von-Mises Stress,

Select >> Fringe >> Quantity >> Stress >> Von-Mises

Display options >> Select continuous tone, deformed, show element edges, animate, frames up to 24 >> OK and show.

14

To Display Displacement :-

Select >> Fringe >> Quantity >> Displacement >> Magnitude

Display options >> Select continuous tone, deformed, show element edges, animate, frames up to 24 >> Ok and show

15

For checking convergence of the results, two graphs were obtained namely, Von-Mises convergence and Strain

Energy convergence. For 10% convergence criterion, the

Strain Energy curve as well as Von-Mises curve didn’t converged. Therefore, another trial run was conducted for

15% as the convergence criterion. Still, convergence was not achieved. As a result, we tried to refine the mesh with in the AutoGEM settings. We changed Edge angle to Max: -

150 and Min: - 30. We also kept the aspect ration as default

30. Even after these AutoGEM changes the analysis didn’t proceed. We therefore utilized the results obtained with

10% convergence criterion.

The steps involved are listed as below:

To display Von-Mises convergence graph :-

Select graph >> Measure >> Select max-stress-vm from pull down list >>Accept >> Ok and show

16

To display Strain energy convergence graph:-

Select graph >> Measure >> Select strain energy from pull down list >> Accept >> Ok and show

The values for different parameters obtained in the results are mentioned below:

1) Maximum Von-Mises Stress = 563.2 N/mm 2

2) Maximum Deflection = 1.863 mm

3) Mass of the structure = 5.743 Kg

4) Factor of Safety = 241 / 563.2 = 0.43

17

Conclusion form graph: - It is seen from the graph that the Von Mises stress increases steadily with each pass with no sign of converging at all, while the strain energy does seem to converge. This behavior coupled with the stress contours indicates that there is a singularity at the junction of the blade and the hub. The P-code method in Mechanica is not able to converge this type of geometry, as it will continue to try to increase the polynomial order indefinitely in order to catch the (theoretically) infinite stress which occurs at the two corners of the blade section.

18

Optimization

Many applications demand certain limiting values put on different parameters of the model without hampering its functionality. The parameters for limiting the values can be geometrical dimensions of the model. At the same time kept goal as optimization of total mass.

When we ran our design study to optimize the mass of the model, we put a limit on the value of Von-Mises stress (241 N / mm

2

) which is the yield strength of the material. Thus the new created design parameters would be such that the maximum Von Mises stresses would not exceed yield strength. The model parameters selected for optimization along with their initial values are listed below:

No

1)

2)

3)

4)

Parameter

Fan Hub Diameter

Fan Hub Thickness

Fan Blade Length

Von-Mises Stress limited to

Initial Value

450 mm

80 mm

276 mm

241 N / mm

2

5) Initial Mass 5.743 Kg

The steps involved in setting up the study are listed below:

Structure >> Struct model >> Design controls >> Design parameters >> Create >> Select the design parameters >> Accept >> Done.

19

Analysis:-

Analysis and design studies >> File >> New design study >> Type- Optimization >>

Limit on max_stress_vm >> Set min, initial and max parameters >> Optim convergence

10% >> Max Iterations 10 >> Accept >> Run the analysis.

20

Results:-

------------------------------------------------------------

Pro/MECHANICA STRUCTURE Version 24.8(819)

Summary for Design Study "fanoptfinal"

Thu Dec 04, 2003 14:10:10

------------------------------------------------------------

Run Settings

Memory allocation for block solver: 128.0

Generate elements automatically.

No errors were found in the model.

Pro/MECHANICA STRUCTURE Model Summary

Principal System of Units: millimeter Newton Second (mmNs)

Length: mm

Force: N

Time: sec

Temperature: C

Model Type: Three Dimensional

Points: 154

Edges: 690

Faces: 944

Springs: 0

Masses: 0

Beams: 0

Shells: 0

Solids: 407

Elements: 407

------------------------------------------------------------

Thu Dec 04, 2003 14:11:35

Goal Minimize: total_mass

Limit: 1

Analysis: finalproject1quickcheck

Load Set: LoadSet1

max_stress_vm < 2.4100e+002

Parameter Min. Value Initial Value Max. Value d34 200 276 300 d2 200 225 300 d1 70 80 90

Optimization Convergence Tolerance: 10 %

Maximum Number of Optimization Iterations: 10

Begin Analysis of Goal and Limits of (14:11:36)

Initial Design

Initial Design Status

Parameters:

d34 276

d2 225

d1 80

Parameters:

d34 276

d2 225

d1 80.2

Goal: 5.7392e-03

Status of Optimization Limits:

1. max_stress_vm 1.2811e+02 < 2.4100e+02 (satisfied)

Resource Check (14:18:23)

Elapsed Time (sec): 495.65

CPU Time (sec): 127.19

21

Memory Usage (kb): 194916

Wrk Dir Dsk Usage (kb): 1

|

|

|

Begin Optimization Iteration 2 (15:03:55)

Begin Line Search (15:05:29)

Step size governed by lower limit on

parameter d34

Line Search Iteration 1

Parameters:

d34 271.399

d2 212.194

d1 70

Goal: 4.6599e-03

Parameters:

d34 200

d2 212.194

d1 71.5722

Goal: 4.5282e-03

Completed Line Search (15:09:10)

Result of Optimization Iteration 2

Parameters:

d34 200

d2 212.194

d1 71.5722

Goal: 4.5282e-03

Status of Optimization Limits:

1. max_stress_vm 1.3250e+02 < 2.4100e+02 (satisfied)

Resource Check (15:20:14)

Elapsed Time (sec): 4206.47

CPU Time (sec): 769.24

Memory Usage (kb): 241117

Wrk Dir Dsk Usage (kb): 0

Begin Optimization Iteration 3 (15:20:14)

Optimization converged on limit boundary.

Best Design Found:

Parameters:

d34 200

d2 212.194

d1 71.5722

Goal: 4.5282e-03

Number of Base Analyses: 6

Number of Perturbation Analyses: 9

-----------------------------------------------------------

Total Elapsed Time (seconds): 4305.30

Total CPU Time (seconds): 794.37

Total Elapsed Time in Parameter Updates (seconds):3619.42

Total Engine Elapsed Time Minus Param. Updates (seconds):685.88

Total CPU Time in Parameter Updates (seconds): 348.57

Total Engine CPU Time Minus Param. Updates (seconds):445.80

------------------------------------------------------------

Run Completed

Thu Dec 04, 2003 15:21:53

------------------------------------------------------------

22

It took one hour and twenty minutes to get the results for this optimization run.

Optimized values of the parameters are listed as below:-

No

1)

2)

3)

4)

5)

6)

Parameter

Fan Hub Diameter

Fan Hub Thickness

Fan Blade Length

Von-Mises Stress

Factor of Safety

Optimized Mass

Optimized Value ( Least Mass)

424.38 mm

71.572 mm

200 mm

132.5 N / mm 2

1.88

4.528 Kg

After optimization we obtained Factor of safety as 1.88 (1.5 < 1.88 < 2) which is desirable. We also came up with reduced mass as 4.528 Kg.

23

References:

1) "Machine Design, An Integrated Approach" By Robert L. Norton, Appendix C, 1996.

2) NACA website: http://www.aae.uiuc.edu/m-selig/ads.html

3) http://www.aeronet.co.kr

4) http://www. moorefans.com

5) http://aero.hanyang.ac.kr

24

Download